How Many Samples Are Required in Geotechnical Site Investigation?

Site investigation is an important part of a geotechnical engineering project. In practice, only a small portion (e.g., 1/100,000 or less) of the total volume of soil is examined, resulting in the significant uncertainty in the interpreted ground conditions. The interpretation error is significantly affected by the number of measurement data (i.e., sample size). Generally speaking, a large sample size provides detail information about the ground conditions, leading to increasing reliability of the interpreted ground conditions and reduced risk to engineering projects. However, increasing sample size requires increasing investigation cost, time and human resources. Therefore, it is important to develop a rational method for determining the required sample size which can provide a satisfactory level of accuracy for the interpreted geotechnical properties profiles. Sampling theory is a well-known topic in statistics, and it can be found in many textbooks on statistics. However, traditional sampling theory only deals with independent data and has limited application in geotechnical site investigation, because geotechnical properties are usually spatially varying and correlated. There is no scientific or quantitative method available for sample size determination in geotechnical site investigation considering spatial variation and correlation of geotechnical properties in existing design codes around world. To address these issues, this paper presents a rational method and statistical charts for sample size determination with consideration of spatial variation and correlation using Bayesian compressive sensing or sampling. Real laboratory test data is used to illustrate application of the statistical chart, and the method is shown to perform well.